摘要
临床辅助诊断对医学图像视觉效果提出了较高的要求,但非下采样轮廓波变换(NSCT)分解获得的低频子带系数不具有稀疏性,不利于保持源图像的细节信息。针对上述问题,提出了一种结合稀疏表示与脉冲耦合神经网络(PCNN)的医学图像融合算法。首先,通过NSCT将源图像分解成低、高频子带系数;其次,利用K奇异值分解(K-SVD)法训练低频子带系数获取过完备字典D,利用正交匹配追踪(OMP)算法稀疏编码低频子带系数,完成低频子带稀疏系数的融合。然后,利用高频子带的空间频率激励脉冲耦合神经网络,根据点火次数选择高频子带的融合系数。最后,将融合的低、高频子带系数通过NSCT逆变换重构出融合的医学图像。实验结果表明:对于边缘信息传递因子Q^(AB/F)指标,该算法灰度和彩色图像融合结果相比于对比算法约提升了34%和10%,且融合结果综合性能优于现有算法。
The clinical auxiliary diagnosis needs a higher requirement for the visual effects of medical images,but the low frequency subband coefficients obtained by the non-subsampled contourlet transform( NSCT)decomposition were not sparse and not conducive to maintain the details of the source image. So a medical image fusion algorithm combining sparse representation and pulse coupled neural network( PCNN) was proposed.Firstly,the original image was decomposited by non-subsampled contourlet transform( NSCT) to obtain low and high frequency subbands ceefficients. Secondly,the K singular value decomposition( K-SVD)method was used to train the low frequency subband coefficients to obtain the over-complete dictionary matrix D. The orthogonal matching pursuit( OMP) algorithm was used to encode the low frequency subband coefficients,which achieved the fusion of sparse coefficients of the low frequency subband. Then,the spatial frequency of high frequency subband was used to excitate PCNN.The coefficient of the larger ignition frequency was selected as the fusion coefficient of high frequency subband. Finally,the NSCT inverse transform was applied to low and high frequency subband fusion coefficients to obtain the fused medical image. The experimental results show that the gray and color image fusion results of the proposed algorithm rise by 34%and 10% than the contrast algorithm in the edge information transfer factor Q^(AB/F) index. The comprehensive performance is superior to the existing algorithm.
出处
《河南科技大学学报（自然科学版）》
CAS
北大核心
2018年第2期40-47,共8页
Journal of Henan University of Science And Technology:Natural Science
基金
国家自然科学基金项目(41505017)
